Open-weights and open-source model releases, Hugging Face trends, local models and self-hosting. 71 stories in today’s issue from 16 outlets, ranked by heat.
From issue #006, November 6, 2026. Filter the live feed.
image-text-to-text · license mit
Chris Alexiuk has a simple rule about open source AI: If you believe in it, be loud about it. “We should be shouting about this technology from the rooftops,” said Alexiuk, a Toronto-based senior product research engineer in deep learning at NVIDIA who…
I’m an MSc CS student with a background in 3D computer vision, trying to figure out how to build visibility and credibility in the AI/ML community. I’ve been thinking about contributing to established open source…
World's first open-source, agentic video production system. 12 production pipelines, 100+ tools, 700+ agent skill and production-knowledge files. Turn your AI coding assistant into a full video production studio.
image-to-3d · license apache-2.0
image-text-to-text · license other
Handbook: https://huggingface.co/spaces/huggingface/moe-blogpost Hello, I'm Nouamane, one of the authors. I'm at Hugging Face and also worked on the Ultra-Scale and Smol Training Playbooks, if you've read those. This…
🚀 PR Agent: The Original Open-Source PR Reviewer. This project is not the Qodo free tier.
Baidu releases ERNIE 5.5 Flash Base: a 280-billion-parameter MoE model pre-trained on 33.5 trillion tokens, shipped deliberately without fine-tuning, so developers can take it wherever they want. The article Baidu releases ERNIE 5.5 Flash Base with 280…
image-text-to-text · license apache-2.0
image-text-to-text · license other
Kimi K3.1 and Tencent’s Hy4 are both gutting out cybersecurity, reverse engineering, and vulnerability-patching capabilities from their open-source models. This is becoming a really bad pattern across these latest…
Model Context Protocol server for Roblox Studio. Connect AI assistants (Claude, Cursor) to read object trees, edit scripts, and run playtests safely. Open-source and lightweight
Make webcam hand games in the browser, or point your coding agent here and have it make one. MediaPipe hand tracking + a gesture engine + a starter game + a playbook.
text-generation · license other
Uni-robot-agent enables open-weight robot-use agents to surpass proprietary models in zero-shot manipulation and suggests joint development of four core capabilities for future training.
Looks like a new super resolution and their samples look promising. https://huggingface.co/kopper/RelayVSR/tree/main
Ricursive Intelligence just released Ripple-65B Preview . Ripple-65B Preview is a 65B dense open-weight model from Ricursive, pretrained from scratch for agentic coding. It is a general-purpose variant of the proprietary model behind Ricursive’s AI chip…
闪猫造物Flashduck — an open-source ESP32-S3 desktop duck robot with six servos, a single animated eye, wide-stance walking and two interaction systems (voice assistant / DuckClaw local agent).
license apache-2.0
NorthCUA scales computer-use agents via an experience flywheel, recursive self-improvement, staged learning, and reinforcement learning to achieve state-of-the-art open-source GUI and web interaction results.
image-to-video · license other
Hey guys, just published a preprint on arXiv proposing a framework we call Skillware to handle agent reliability. (arXiv:2611.07652v1) The main idea is that instead of using prompt engineering to get models to write…
One app for every document. Built for AI agents. An open-source desktop office suite written in Rust: documents, spreadsheets, slides, PDF, Markdown, code and rich text, opened, edited and saved in one row of file tabs. The same binary is…
Curated list of self-hosted open-source AI: local LLMs, chat UIs, agents, MCP servers, RAG, speech and image generation. Health-checked weekly, dead projects removed automatically.
automatic-speech-recognition · license other
The weights of Mistral Large 4 (Le Chonk) are now open. A 1T parameter multimodal model, and best open-weight model for coding, trained in Europe.
image-to-video · license other
license apache-2.0
I'm the developer of Kokon, a local photo editor around Qwen-Image 2.1 Turbo (on stable-diffusion.cpp, and on its own engine on Macs). I posted it here a few weeks ago. 0.12 adds a Layers lab: click objects in any…
image-text-to-text · license other
image-to-image · license other
35 more stories, by section and heat.
The hottest Open-weights AI stories from the previous 5 issues.